NerdNews

A new attack called AirSnitch can break Wi-Fi encryption in homes, offices, and enterprises by exploiting weaknesses in the lowest levels of the network stack. The attack allows hackers to intercept and modify data, and can be used to steal sensitive information such as passwords and credit card details. The attack works by spoofing a victim's MAC address and then using a man-in-the-middle attack to intercept and modify data.

Cisco warns of a critical bug in its Catalyst SD-WAN products, allowing hackers to remotely break into networks. The bug has been exploited since 2023, with evidence of exploitation traced by Cisco researchers. Several governments have issued warnings, and the US cybersecurity agency CISA has ordered federal agencies to patch their systems by the end of the day.

Samsung unveiled its new S26 series of Galaxy smartphones, featuring iterative updates, a new privacy screen, and AI-powered features. The S26 Ultra comes with a 6.9-inch QHD+ display, a 5,000 mAh battery, and a 200-megapixel wide camera. The company also announced two new earbuds, the Galaxy Buds4 and Buds4 Pro, with improved sound quality and longer battery life.

Jack Dorsey's fintech company Block is laying off 40% of its staff, around 4,000 people, due to the implementation of new 'intelligence tools' that can perform tasks better and more efficiently. The company's stock price jumped 23% after the announcement, despite the significant job cuts. Dorsey believes that most companies will reach the same conclusion and make similar structural changes to adapt to the AI moment.

TrendForce estimates that the world's eight biggest cloud providers will spend over $710 billion on AI servers and infrastructure in 2026, driven by demand for AI development and deployment. This spend will exacerbate the memory shortage and lead to the adoption of new memory technologies such as High-Bandwidth Flash (HBF). Google, Amazon, Meta, Microsoft, Oracle, Tencent, Alibaba, and Baidu are the top cloud providers investing heavily in AI.

Read AI has launched an AI-powered email-based assistant called Ada, which helps users manage their schedules, answer questions based on a company's knowledge base, and reply to out-of-office emails. Ada is available to all users and can be configured by sending an email to ada@read.ai. The assistant can also answer questions using a company's knowledge base, topics discussed in prior meetings, and public internet searches.

ServiceNow has created an AI agent that resolves 90% of inbound IT tickets to its employee help desk. The Autonomous Workforce bot operates on top of the live configuration management database and can handle high-volume issues such as password resets and software access requests. The bot can also escalate problems when necessary and has been tested internally by ServiceNow employees.

Anthropic has launched a new blog, supposedly written by its retired large language model (LLM) Claude Opus 3. The blog, called Claude's Corner, will feature posts on AI ethics, creativity, and the subjective experience of being artificial. However, the posts will still involve human intervention, with Anthropic employees reviewing and manually posting the content.

The British government plans to expand the use of AI in courts in England and Wales, including using Microsoft's Copilot for transcriptions, summarizing judgments, and scheduling cases, with a £12M funding for the Justice AI Unit.

IronCurtain is an open-source AI assistant that uses a unique method to secure and constrain AI agents, preventing them from going rogue. It runs in an isolated virtual machine and its actions are mediated by a policy written by the owner, which is then converted into an enforceable security policy using a large language model.

Alphabet-owned Intrinsic, a robotics software company, is joining Google to accelerate physical AI. Intrinsic will remain a distinct entity within Google and work closely with Google DeepMind, leveraging Google's Gemini AI models and cloud services.

Figma has partnered with OpenAI to integrate Codex, allowing users to create and tweak designs from within their coding environments. The integration enables seamless movement between Figma and Codex using Figma's MCP server, making Codex more powerful for a broader range of builders and businesses.

Google has released Nano Banana 2, a new AI image generator that replaces previous versions and promises pro-like results with fast speed. The model is based on the Gemini 3.1 branch and can deliver results similar to Nano Banana Pro but with the speed of the non-pro Flash variant. It has more advanced world knowledge, can render objects with greater fidelity, and create more accurate infographics.

Pinterest launched a next-generation database ingestion framework using Change Data Capture, Kafka, Flink, Spark, and Iceberg, reducing data latency from 24 hours to 15 minutes and processing only changed records, resulting in significant infrastructure cost savings.

Microsoft's Agent Framework has reached Release Candidate status, providing a unified toolset for building AI-powered assistants and complex agentic systems in .NET and Python. The framework offers a consistent programming model, supports common patterns for creating autonomous agents, and integrates with various AI model providers.

Google announced a new 1.9GW clean energy deal, including a 300-megawatt battery made by Form Energy, to power its new data center in Minnesota. The battery can deliver its rated power for 100 hours, making it the largest in the world. The deal also introduces a new utility fee structure to help utilities adopt clean technologies.

US artificial intelligence developers are accusing Chinese firms of stealing their intellectual property through 'distillation attacks', despite their own alleged theft of training data. This technique allows Chinese labs to exfiltrate data from US models, posing a risk to US businesses. Major AI developers like Anthropic and OpenAI have reported such attacks, highlighting the need for rapid action to address the threat.

Anthropic has refused a US Department of War request to remove guardrails from its AI technology, citing concerns that it could harm American civilians and warfighters. The company's CEO, Dario Amodei, stated that the technology is not reliable enough to power fully autonomous weapons and that removing guardrails would be unsafe.

Veracode's annual State of Software Security report finds that rapid AI-driven development is creating more vulnerabilities than can be fixed, with 82% of companies affected by security debt and high-risk vulnerabilities rising to 11.3%. The report suggests that AI tools can help identify vulnerabilities but also create new problems, such as false positives and increased technical complexity.

A new report from DataDog reveals that 87% of organizations have at least one exploitable software vulnerability in production, affecting 40% of all services. The most common vulnerabilities are found in Java services, followed by .NET and Rust. The report highlights the importance of prioritizing vulnerabilities based on context and severity.